Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Enlisting the support of trusted sources to tackle policy problems: The case of antimicrobial resistance

  • Aaron Martin ,

    Roles Conceptualization, Investigation, Methodology, Project administration, Writing – original draft, Writing – review & editing

    Affiliation School of Social and Political Sciences, Faculty of Arts, University of Melbourne, Victoria, Australia

  • Timothy B. Gravelle,

    Roles Conceptualization, Data curation, Formal analysis, Methodology, Writing – original draft, Writing – review & editing

    Affiliation Institute for Social Science Research, University of Queensland, Queensland, Australia

  • Erik Baekkeskov,

    Roles Conceptualization, Writing – original draft, Writing – review & editing

    Affiliation School of Social and Political Sciences, Faculty of Arts, University of Melbourne, Victoria, Australia

  • Jenny Lewis,

    Roles Conceptualization, Methodology, Writing – original draft, Writing – review & editing

    Affiliation School of Social and Political Sciences, Faculty of Arts, University of Melbourne, Victoria, Australia

  • Yoshi Kashima

    Roles Conceptualization, Methodology, Writing – review & editing

    Affiliation Melbourne School of Psychological Sciences, Faculty of Medicine, Dentistry and Health Sciences, University of Melbourne, Victoria, Australia


Antimicrobial resistance represents one of the world’s most pressing public health problems. Governments around the world have—and will continue to—develop policy proposals to deal with this problem. However, the capacity of government will be constrained by very low levels of trust in government. This stands in contrast to ‘medical scientists’ who are highly trusted by the public. This article tests to what extent trusted sources can alter attitudes towards a policy proposal to regulate the use of antibiotics. We find that respondents are much more likely to support a policy put forward by ‘medical scientists.’ This article provides some initial evidence that medical scientists could be used to gain support for policies to tackle pressing policy challenges such as AMR.


Antimicrobial resistance (AMR) is now recognized as one of the world’s most pressing public health problems [1,2]. Governments around the world have—and will continue to—develop policy proposals to deal with this problem. However, the capacity of government will be constrained by levels of trust in government. Trust in government has been declining around the world in recent years [36]. Kettl [7] argues that ‘the rising tide of distrust in government is surely one of the biggest challenges facing the world’s democracies in the twenty-first century.’ In such an environment, government may find it harder to secure support for policy proposals to deal with AMR. At a time when trust in government is declining and social problems (such as AMR) are becoming more complex, it is likely that government will have to enlist the support of trusted sources to secure support for policy reforms. This article tests to what extent trusted sources can alter attitudes toward a policy proposal to regulate the use of antibiotics.

Trusted sources and AMR

Trust in government and scientists

While it is clear that trust in government in many countries is at historically low levels [7,8], a less remarked-upon pattern is that not all authorities are distrusted to the extent that government is. This is especially true when we compare the authorities manipulated in this experiment: government versus scientists (as explained in the methods section below). Public opinion data from the United States shows that ‘medical scientists’ are ranked as the most trustworthy of all professions [9]. These findings have been replicated cross-nationally [10]. Taken together this research provides very strong evidence that while scientists are highly trusted, government is not.

The differences between trust in government and trust in scientists could have important consequences. Source effects have been of long-standing interest across the social sciences [11]. Sources have been found to have the largest effect on attitudes when they are considered credible and trustworthy [1214]. In policy contexts we would expect little-trusted politicians to be a less effective source for policy messaging than other sources, such as ‘scientists’. Given that trust varies by source, the way source shapes support for policy is an important question. Therefore, we expect that citizens will be more likely to support a policy proposal put forward by a trusted source. Because source effects are complex [15] it is important to investigate this experimentally for the particular case of AMR.

We test this hypothesis using a unique survey experiment fielded in Australia. While we limit our study to one country, we believe this case is instructive for a number of reasons. First, Australia has experienced the same steep declines in political trust experienced in other western democracies [6]. Second, we also find the same patterns in Australia in terms of scientists being seen as trustworthy and contributing positively to society–much more so than government [16,17]. Australia therefore provides a useful test case for the broader phenomenon where trust in government is low (and declining), but scientists are highly trusted.

Antimicrobial resistance

Antibiotic and other antimicrobial resistance (AMR) has been known since antibiotics were discovered in the 1940s [18] Yet it is now a rapidly emerging area for public policy-making, in Australia and globally [1,19,2]. Species of bacteria and other microbes evolve under the selection pressures imposed by antibiotics and other antimicrobials to become resistant to such interventions. In turn, evolution renders these mainstays of human and veterinary medicine less effective. AMR also has implications for food safety. As animals and crops raised in close proximity become less protectable from the spread of microbials, contemporary intensive mass production of foodstuffs becomes less viable. The Australian government is one of many across the globe recognising that ‘the appropriate and judicious use of antimicrobials is essential to slowing the emergence of resistance’ of microbes to antimicrobials [20,1]. In particular, limiting uses of antibiotics to the most necessary medical applications is one of the key public policies for curbing AMR.

In our experiment, we measure support for a policy proposal to regulate and monitor the use of antibiotics. AMR is an interesting policy area because we would expect ex-ante that citizens would be supportive of any plausible attempt to curb it. In this sense the attitudes towards curbing AMR may be regarded as a ‘hard-test’ of the expectation that the public will support a policy reform put forward by scientists. It is not a politically contested issue like immigration or climate change [2124], nor very controversial.

AMR has received repeated attention from scholars at the intersection of social science and medicine in the last two decades [2527]. Prior policy-related research has looked at factors diverse as pharmacists’ administering of antibiotics [28], antibiotics options markets [27], and antibiotic use in low income countries [25]. Even more specific to our focus are surveys and more recently citizen juries that have been used to identify drivers of public demand [2931] and of physicians’ prescribing patterns [3233]. Experimental and observational studies of public information impact on uses of antibiotics have identified that information campaigns can limit popular demand for or physicians’ propensity to prescribe antibiotics [34,35]; yet there is little clarity about ‘which [information campaign] strategy is most efficient in changing attitudes and practices’ [36].

Examining the effect of trusted sources

No research (that we are aware of) has been done to test the effect of enlisting trusted sources to gain support for a policy to curb AMR. We hypothesise that one consequence of declining support for government may be that citizens are more likely to support a policy promoted by a ‘non-political actor,’ rather than government. For AMR in particular, we expect a policy proposal put forward by ‘medical scientists’ will garner more support than a policy proposal put forward by ‘government.’ Beyond this, we also test how policy support is shaped by the organisation’s (whether ‘medical scientists’ or ‘government’) perceived competence in delivering effective policy. Literature has also shown that competence can have an effect on political trust [5] which may in turn have an effect on support for policy. We examine this by sending signals about the level of competence of the two different sources.

Based on this literature we tested two hypotheses:

  1. H1: A policy proposal from a trusted source (medical scientists) will receive higher support than a proposal from a less trusted source (the government in Canberra).
  2. H2: A policy proposal coming from a source that is signalled to be competent will receive higher support.

Data and methods

To test these theoretical expectations, we designed an online survey experiment, which we administered to a demographically representative non-probability sample of the Australian public (n = 388) through Qualtrics, an online survey platform. The sample (derived from an online panel of respondents supplied by Qualtrics) was balanced on age, gender, educational attainment, and party identification (the latter was matched to 2016 Australian Election Study data). This study was approved by the Human Ethics Committee at the University of Melbourne, Australia.

The use online non-probability samples in quantitative social science has expanded substantially in the past decade. While some studies find that probability and non-probability samples yield different estimates [37,38], there is growing evidence that there are few meaningful differences between the inferences obtained from online non-probability panels and probability-based samples [3941]. Our experimental design also means that random assignment is far more important than random selection. Our sample is thus fit for purpose [42].

The experiment was designed as a 2 x 2 factorial experiment involving two treatments: (1) a federal government/medical scientist framing of the policy proposal, and (2) a competence frame highlighting the policy expertise and effectiveness of the federal government/medical scientists (versus a control condition where nothing about competence is presented to the respondent). Full details of the experiment (including the full survey questionnaire) are provided in the appendix.

After a generic description of the dangers and causes of anti-microbial resistance, respondents were presented with a policy proposal from either ‘the government in Canberra’ or ‘Australia’s leading medical scientists.’ Specifically, respondents were told that:

[The government in Canberra/Australia’s leading medical scientists] believe that an effective way to address this problem is to more closely monitor and regulate how medical professionals prescribe antibiotics. Accordingly, [the government in Canberra/Australia’s leading medical scientists] have put forward a proposal whereby they would monitor how many antibiotics medical professionals are prescribing and, where necessary, enforce that they prescribe less.

Respondents assigned to receive the competence treatment were further told that:

[The government in Canberra/Australia’s leading medical scientists] has a long and strong record of successfully delivering programs like this. For example, in 2016 they successfully monitored a program which asked surgeons to fill out a checklist after every procedure to ensure no mistakes were made. This was estimated to have saved 23,000 lives and prevented over 40,000 infections. This track record shows that the program is likely to be a success.

We then presented respondents with a series of comprehension checks to confirm that respondents had read the article properly–based on which we deleted 12 observations from those who failed to correctly identify what topic (i.e. AMR) the article referred to; all respondents correctly answered the second manipulation check asking who the actor advocating for the policy (i.e. government versus medical scientists) was. Respondents were then asked about our key outcome measure: ‘Do you support the proposal for [the government in Canberra/Australia’s leading medical scientists] to monitor how many antibiotics medical professionals are prescribing and, where necessary, enforce that they prescribe less?’ Response categories were a four-point scale from ‘Strongly agree’ to ‘Strongly disagree.’

For our statistical analysis, we first examine the distribution of responses across the two experimental conditions in simple bivariate tables. Given the four ordered categories of the dependent variable, we then model the effects of the trusted source and competence frames using ordinal logistic regression [43]. The ordinal logit coefficients are thus our estimates of the effects of the two experimentally-manipulated frames on the cumulative log odds of providing a higher response on the dependent variable (i.e., expressing greater support for the proposal), controlling for party identification. For ease of interpretation, the effects of the two experimental frames obtained from the ordinal logistic regression results are also reported as predicted probabilities [44].


Examining first the bivariate distributions for support for prescription monitoring by each treatment, it is clear that the proposal enjoys majority support regardless of how it is framed. Overall support (for the entire sample) is 91.8 percent (see Table 1). Still, the framing of the proposal does strengthen or weaken the level of support. Framing the proposal as coming from the government reduces aggregate support to 85.1 percent (versus 98.5 percent when the proposal is framed as coming from Australia’s leading medical scientists). The modal response also shifts from supporting the proposal ‘strongly’ when presented as originating with Australia’s leading medical scientists to weaker (‘somewhat’) support when presented as a government proposal. By contrast, the competence frame increases support to 94.8 percent (versus 88.7 percent when no competence frame is presented). The modal response similarly shifts from ‘somewhat’ to ‘strongly’ supporting the proposal with the competence frame.

Table 1. Support for monitoring the use of antibiotic monitoring.

It is notable that while the government frame reduces support for antibiotic monitoring, highlighting organisational competence appears to (at least partly) counteract this negative effect. To reconfirm this result, and to understand how the two frames work in combination with each other (and with other factors, such as party identification), multivariate analysis is required. The results from the ordinal logit models indicate that both the government frame and competence frame continue to exert statistically significant effects when evaluated in tandem (see Table 2). The government frame (contrasted to the medical scientist frame) reduces the cumulative log odds of supporting the antibiotic monitoring proposal by 1.42 (95% c.i.: -1.84 to -0.99). As suggested above, this negative effect is partly (though not completely) mitigated by the competence frame, which increases the cumulative log odds of support by 0.84 (95% c.i.: 0.43 to 1.25). Incidentally, additional analyses not reported here fail to find a significant interaction between the government and competence frames–that is, there is no evidence that the competence frame exerts a different effect when combined with the government frame or the medical scientist frame. While we may expect those who identify with the Liberal Party (the current party in power) to have greater support for the policy proposal we find that party identification has no effect on the responses.

Table 2. Explaining support for antibiotic monitoring (ordinal logit).

The results of the regression analysis are easier to interpret when translated into predicted probabilities on a 0–1 scale (see Table 3). Presented in this way, the results reconfirm those from Table 1 and Table 2. Specifically, framing the proposal as coming from the government (as opposed to medical scientists) reduces the predicted probability of strongly supporting the proposal by 0.34. By contrast, the competence frame raises the predicted probability of strongly supporting the proposal by 0.20. In short, drawing attention to the competence of the source supporting the antibiotic monitoring proposal does not fully make up for the loss of support when it comes from the government.

Table 3. Predicted probabilities of support for antibiotic monitoring.


Political trust is important because it has been shown to have numerous attitudinal and behavioural consequences [4,5,45]. In this article we have explored one such consequence: whether an AMR policy proposal put forward by scientists is more likely to be supported than one put forward by government. As we noted above, the experiment was in a lot of ways a ‘hard-test’ given that most people might recognise AMR as a pressing public health problem and be supportive of any attempt to curb it. This is borne out, with high aggregate levels of support for the policy. Even allowing for that, we find a significantly higher level of support for the policy proposal put forward by scientists, as opposed to government.

What mechanisms may be driving this result? While we have not been able to test this in the current study, it seems likely that the term ‘medial scientists’ triggers a number of positive associations in the mind of the public, whereas ‘government’ triggers a number of negative associations. Medical scientists also hold a special status in health policy—medical expertise is important, strongly legitimised through institutions, and embedded throughout policy networks [46]. One might expect the medical scientists to be more trusted by the public because of this. We have some additional evidence that trust is in fact the mediating factor with trust in the effectiveness of the actor in administering this policy (which is a partial proxy for general trust in the actor) being much higher in the medical scientist condition as opposed to the government condition. (In our survey asked the following post-treatment question: How much confidence do you have in the [federal government in Canberra/Australia’s leading medical scientists] to EFFECTIVELY monitor and enforce the use of antibiotics?’ While 49 percent of respondents receiving the ‘Australia’s leading medical scientists’ condition expressed ‘a lot of trust’, only 13 percent of respondents receiving the ‘federal government in Canberra’ condition responded in the same way (χ2 = 79.59, d.f. = 3; p < 0.001). This, of course, is not general trust and is conditioned on treatment. Still, this provides additional evidence of the potential causal mechanism). The results support our hypothesis (and the literature on source effects) that policy proposals put forward by trusted sources are much more likely to be supported. Future research ought to test trustworthiness of particular actors as a mediating variable.

Larger questions remain that we think are fruitful avenues for future research. For example, while we expect the findings here to be replicated in other settings where government is distrusted and scientists are trusted, this remains a hypothesis until we have data to confirm this. Coordination of a policy regulating antibiotic use is also complicated by the number of individuals involved in administering antibiotics—Broom, Broom Kirby and Scambler [26] show that antibiotic use is an interprofessional rather than a purely medical problem. There are also broader normative questions about whether it is desirable for medical scientists to be engaged in such policy advocacy and implementation.

There are of course several limitations to this study. This study employs a medium-sized sample in a particular setting. The effects we find may vary across issue and country. For example, we don’t know what happens when respondents are confronted with more controversial policies concerning issues such as immigration or climate change where respondents may have strong priors or take cues from party leaders. We have also not tested what happens to the reputation of scientists if and when they enter political debate. That said, for an uncontroversial issue like AMR there may not be much reputational damage.

Even allowing for these limitations this article has provided credible evidence that medical scientists could be used to gain support for policies that will be essential if AMR is to be curbed. This article shows that the public are much more likely to support a policy proposal to deal with AMR put forward by ‘scientists’ as opposed to ‘government.’ It also shows that drawing attention to the competence of medical scientists or government alters attitudes towards the policy proposal. In an environment of growing distrust in government alongside pressing public health problems such as AMR it is essential that such questions are explored.


This research was supported by grants from the Australian Research Council (Understanding the Causes of Political Trust, DE160100603) and the Policy Lab, University of Melbourne. Aaron Martin would like to thank Marc Hetherington and Cindy Kam for feedback at the early stages of this research project and the Center for the Study of Democratic Institutions at Vanderbilt University for hosting a visit that allowed some of this research to be completed.


  1. 1. WHO. Global Action Plan on Antimicrobial Resistance. 2015. Available from:; cited June 2018.
  2. 2. UN. At UN, global leaders commit to act on antimicrobial resistance: collective effort to address a challenge to health, food security, and development, Press release: High-Level Meeting on Antimicrobial Resistance, General Assembly of the United Nations, 21 September 2016. Available from: Cited 12 June 2018.
  3. 3. Pharr SJ, Putnam RD editors. Disaffected Democracies: What's Troubling the Trlilateral Countries? Princeton, NJ: Princeton University Press; 2000.
  4. 4. Dalton RJ. Citizen Politics: Public Opinion and Political Parties in Advanced Industrial Countries. New York: Chatham House; 2002.
  5. 5. Hetherington MJ. Why Trust Matters: Declining Political Trust and the Demise of American Liberalism. Princeton University Press: New York; 2005.
  6. 6. McAllister I, Cameron S. Trends in Australian political opinion: Results from the Australian election survey 1987–2016. Canberra: ANU; 2016.
  7. 7. Kettl DF. Can Governments Earn Our Trust? New York: Polity Press; 2017.
  8. 8. Mounk Y. The People Vs. Democracy: Why Our Freedom is in Danger and How to Save It. Cambridge, MA: Harvard University Press; 2018.
  9. 9. Funk C, Kennedy B. Public confidence in scientists has remained stable for decades. Pew Research Center. 2018. Available from: Cited June 2017.
  10. 10. Ipsos MORI. Veracity Index 2017 Ipsos MORI. Available from: Cited 26 June 2018.
  11. 11. Wilson EJ, Sherrell DL. Source Effects in Communication and Persuasion Research: A Meta-Analysis of Effect Size. Journal of the Academy of Marketing Science. 1993; 21(2): 101–112.
  12. 12. Lupia A. Who can persuade whom? Implications from the nexus of psychology and rational choice theory. In ed. James H. Kuklinski JA, editor. Thinking about political psychology. New York: Cambridge University Press. 51–88; 2002.
  13. 13. Lupia A, McCubbins M. The democratic dilemma. Cambridge: Cambridge University Press; 1998.
  14. 14. Petty RE, Wegener DT. Attitude change: Multiple roles of persuasion variables. In Gilbert DT, Fiske ST, Lindzey G editors. The handbook of social psychology. Boston: McGraw-Hill. 323–390; 1998.
  15. 15. Pornpitakpan C. The persuasiveness of source credibility: A critical review of five decades’ evidence. Journal of applied social psychology. 2014; 34(2), 243–281.
  16. 16. Searle SD. How do Australians engage with science? Preliminary results from a national survey. Australian National Centre for the Public Awareness of Science (CPAS). Canberra: The Australian National University; 2014.
  17. 17. Lamberts R. The Australian Beliefs and Attitudes Towards Science Survey. The Australian National University. Canberra, Australia; 2017
  18. 18. Wallinga D, Rayner G, Lang T. Antimicrobial resistance and biological governance: explanations for policy failure. Public Health; 2015; 129(10), 1314–1325. pmid:26454427
  19. 19. Carlet J, Pulcini C, Piddock LJ. Antibiotic resistance: a geopolitical issue. Clinical Microbiology and Infection. 2014; 20(10), 949–953. pmid:25040923
  20. 20. Australian Government (AG) Responding to the threat of antimicrobial resistance: Australia’s first national Antimicrobial Resistance strategy 2015–2019. 2015. Available from Cited 12 June 2018.
  21. 21. Bilodeau A., and Fadol N. The Roots of Contemporary Attitudes toward Immigration in Australia: Contextual and Individual-Level Influences. Ethnic and Racial Studies. 2011; 34(6): 1088–1109.
  22. 22. McCright AM, Dunlap RE. The politicization of climate change and polarization in the American public’s views of global warming, 2001–2010. The Sociological Quarterly. 2011; 52: 155–194.
  23. 23. Egan PJ. Partisan Priorities: How Issue Ownership Drives and Distorts American Politics. Cambridge: Cambridge University Press.; 2012.
  24. 24. Tranter B. The Great Divide: Political Candidate and Voter Polarisation over Global Warming in Australia. Australian Journal of Politics and History. 2013; 59(3): 397–413.
  25. 25. Radyowijati A, Hilbrand H. Improving antibiotic use in low-income countries: an overview of evidence on determinants. Social Science and Medicine. 2013. 57(4); 733–744.
  26. 26. Broom A, Broom J, Kirby E, Scambler G. The path of least resistance? Jurisdictions, responsibility and professional asymmetries in pharmacists' accounts of antibiotic decisions in hospitals. Social Science and Medicine. 2015; 146; 95–103. pmid:26513118
  27. 27. Brogan DM. and Mossialos E. A critical analysis of the review on antimicrobial resistance report and the infectious disease financing facility. Globalization and Health. 2016: 12(1). ISSN 1744-8603 pmid:27000847
  28. 28. Broom A, Broom J, Kirby E. Cultures of Resistance? A Bourdieusian Analysis of Doctors’ Antibiotic Prescribing. Social Science and Medicine. 2014; 110:81–8. pmid:24727665
  29. 29. McNulty CAM, Boyle P, Nichols T, Clappison P, Davey P. The public’s attitudes to and compliance with antibiotics. Journal of Antimicrobial Chemotherapy. 2007; 60(suppl. 1): i63–i68.
  30. 30. Wang L, Zhang X, Liang X. Bloom G. Addressing antimicrobial resistance in China: policy implementation in a complex context. Globalization and Health. 2016; pmid:27267876
  31. 31. Degeling C, Iredell J, Nguyen K, Norris JM, Turnidge JD, Dawson A, Gilbert GL. Assessing the public acceptability of proposed policy interventions to reduce the misuse of antibiotics in Australia: A report on two community juries. Health Expectations; 2018; (21): 90–99.
  32. 32. Trubiano J, Chen C, Cheng A, Grayson M, Slavin M,Thursky K. Antimicrobial allergy 'labels' drive inappropriate antimicrobial prescribing: lessons for stewardship. Journal of Antimicrobial Chemotherapy. 2016; 71(6): 1715–22. pmid:26895771
  33. 33. Weier N, Thursky K, Zaidi STR. Antimicrobial knowledge and confidence amongst final year medical students in Australia. PLoS One. 2017; 12(8), e0182460. pmid:28771549
  34. 34. Gonzales R., Steiner JF, Lum A, Barrett PH. Decreasing antibiotic use in ambulatory practice: impact of a multidimensional intervention on the treatment of uncomplicated acute bronchitis in adults. Journal of the American Medical Association. 1999; 281(16): 1512–1519. pmid:10227321
  35. 35. Sabuncu E, David J, Berne`de-Bauduin C, Pepin S, Leroy M, Boelle P, Watier L, Guillemot D. Significant Reduction of Antibiotic Use in the Community after a Nationwide Campaign in France, 2002–2007. PLoS Medicine. 2009; 6(6), e1000084. pmid:19492093
  36. 36. Huttner B, Goosens H, Verheij T, Harbarth S. Characteristics and outcomes of public campaigns aimed at improving the use of antibiotics in outpatients in high-income countries. Lancet Infectious Diseases, 2010; (10): 17–31.
  37. 37. Chang L, Krosnick JA. National Surveys Via RDD Telephone Interviewing Versus the Internet: Comparing Sample Representativeness and Response Quality. Public Opinion Quarterly. 2009; 73(4): 641–78.
  38. 38. Yeager DS, Krosnick JA, Chang L, Javitz HS, Levendusky MS, Simpser A, Wang R. “Comparing the Accuracy of RDD Telephone Surveys and Internet Surveys Conducted with Probability and Non-Probability Samples. Public Opinion Quarterly. 2011; 75(4): 709–47.
  39. 39. Ansolabehere S, and Schaffner BF. Does Survey Mode Still Matter? Findings from a 2010 Multi-Mode Comparison. Political Analysis. 2014; 22(3): 285–303.
  40. 40. Breton C, Cutler F, Lachance S, Mierke-Zatwarnicki A. Telephone versus Online Survey Modes for Election Studies: Comparing Canadian Public Opinion and Vote Choice in the 2015 Federal Election. Canadian Journal of Political Science 2017; 50(4): 1005–36.
  41. 41. Stephenson LB., and Crête J. Studying Political Behavior: A Comparison of Internet and Telephone Surveys. International Journal of Public Opinion Research. 2011; 23(1): 24–55.
  42. 42. Baker R, Brick JM, Bates NA, Battaglia MP, Couper MP, Dever JA et al. Summary Report of the AAPOR Task Force on Non-Probability Sampling. Journal of Survey Statistics and Methodology. 2013; 1(2): 90–105.
  43. 43. Fox J. Applied Regression Analysis and Generalized Linear Models. 2nd ed. Thousand Oaks, CA: Sage; 2008
  44. 44. Scott LJ, Freese J. Regression Models for Categorical Dependent Variables Using Stata. 3rd ed. College Station, TX: Stata Press; 2014.
  45. 45. Hetherington MJ, Rudolph TJ. Why Washington Won’t Work: Polarization, Political Trust, and the Governing Crisis. Chicago: University of Chicago Press; 2015.
  46. 46. Lewis JM. Being around and knowing the players: networks of influence in health policy. Social Science and Medicine. 2006; 62: 2125–2136. pmid:16289737